NEW YORK, Jan 2, 2026, 09:41 ET — Regular session
Nvidia shares slipped in early trading on Friday, down about 0.5% at $186.50, as chip stocks opened the year mixed. Taiwan Semiconductor Manufacturing was up about 1.4%, while the iShares Semiconductor ETF fell about 1.2%.
The moves come as investors reset positions after a choppy finish to 2025 and look for signals on whether last year’s AI-led rally can broaden or re-accelerate. Nvidia was up about 1.4% in premarket trade, Reuters reported, as traders awaited fresh U.S. economic readings and next week’s labor-market data for clues on the Federal Reserve’s rate path.
Why it matters now: Nvidia remains a key barometer for spending on AI infrastructure, where a single headline on demand or supply can ripple across chips, data centers and mega-cap tech.
China sits at the center of that debate, with investors weighing the sales opportunity against regulatory uncertainty and potential knock-on effects for global chip supply.
A Reuters report on Wednesday said Chinese technology companies have ordered more than 2 million of Nvidia’s H200 graphics processing units (GPUs) for 2026, while Nvidia has about 700,000 units in stock. The H200 is part of Nvidia’s Hopper lineup; the company has approached TSMC to ramp up production, with work expected to start in the second quarter, and has priced China variants around $27,000 per chip, the report said. Reuters added that Beijing has not yet approved H200 imports even after the Trump administration recently allowed exports with a 25% fee, and that Chinese officials are weighing conditions including bundling purchases with domestic chips.
Separately, ByteDance plans to spend about 100 billion yuan ($14.29 billion) on Nvidia AI chips in 2026, the South China Morning Post reported, according to Reuters. Reuters said it could not immediately verify the report and noted the plan hinges on whether Nvidia is allowed to sell its H200 GPUs in China.
Supply chain signals are also in focus ahead of the CES trade show in Las Vegas next week. “On HBM4 in particular, customers have even stated that ‘Samsung is back’,” Samsung Electronics co-CEO and chip chief Jun Young-hyun said in a New Year address reviewed by Reuters, as Samsung works to narrow the gap with memory rival SK Hynix and discussed supplying next-generation high-bandwidth memory to Nvidia. Reuters
High-bandwidth memory, or HBM, is a type of stacked memory designed to feed data quickly to GPUs — the chips used to train and run AI models. When HBM is tight, it can limit how fast AI servers ship, even if GPU demand is strong.
That dynamic helps explain why traders are parsing China headlines so closely. Stronger China demand can lift volumes and pricing power, but it can also stretch capacity across foundry output, packaging and memory.
Next up is CES 2026, where Nvidia’s founder and CEO Jensen Huang is scheduled to speak on Monday, Jan. 5, according to Nvidia and CES schedules. Investors will be listening for any fresh product roadmap details and commentary on supply, demand and regional sales mix.
Wall Street’s next major checkpoint is Nvidia’s earnings calendar. The company is slated to report fourth-quarter fiscal 2026 financial results on Feb. 25, according to its investor relations events schedule.
For now, traders are watching for clearer signals on China import approvals, any shifts in export policy, and whether early-2026 sector rotation keeps pressure on high-multiple AI leaders or drives fresh buying back into the group.
